Sampling strategies for sugarcane using either clonal replicates or diverse genotypes can bias the conclusions of RNA-Seq studies

Abstract

A key procedure for ensuring statistical confidence in differential gene expression analyses is to use biological replicates to compare distinct groups. Biological replicates allow the estimation of the residual variation in the gene expression levels among samples of a given experimental condition. In sugarcane, it is possible to obtain an estimate of residual variability at two levels: among samples of distinct genotypes of the same experimental treatment, or clonal replicates of the same genotype. The sequencing costs are often a limitation to leveraging both these levels in the same study, stressing the relevance of efforts to determine an appropriate experimental design. We aim to investigate this question by comparing the transcriptional profiles of young sugarcane culms with different sucrose levels using both sampling strategies. Our results show that clonal replicates provided enough statistical power to identify nearly three times more deferentially expressed genes than the more diverse strategy. However, it resulted in potentially less meaningful biological results, because many of the significant genes were likely related to the particular genotype of choice, rather than representing a common expression profile for the compared groups. This study supports the development of sound experimental designs in new studies regarding differential expression for sugarcane.

Keywords:
Differential gene expression; vegetatively propagated crops; biological replicates; sugarcane transcriptomics; missing samples